operating programs—the traditional focus of evaluation—does not answer some important questions? research is also needed in planning and implementing programs. During the planning phase, there are quest: of need and how to meet those needs. Survey and ethnographic studies can establish the extent and distribution of an educational problem; controlled pi] testing and field tests can determine the effectivenes and feasibility of alternative interventions for relieving the problem; and economic analyses can be us to make cost estimates. Once a program is establishec and operating, there are questions of fiscal and cove] accountability. Analyses of administrative records GC determine whether funds are being used properly and whether the program is reaching the intended beneficiaries, although supplementary fiscal audits ai beneficiary studies are sometimes required. Finding < whether the program is being implemented appropriately requires, in addition to program administrative recorc special surveys of program services and ethnographic studies. Finally, there are questions of program impc they can be addressed definitively only through rigorc and often costly research methods. Consequently impac evaluation should be undertaken only if the requisite skills and resources are available.
Not all programs can be fully evaluated: that is, all questions can be answered for all programs. In particular, meaningful impact evaluation is possible c for programs for which intended beneficiaries and effe can be clearly specified. There are two kinds of programs for which such specification is extremely difficult or impossible. For a program having vague goals or many diverse goals, evaluators and those who commission an evaluation must be able to agree on whic goal should be assessed and whether appropriate measui are available to assess it. For a program in which lc sites are given autonomy to develop their own specific objectives and means of reachina them- one cannot